OWL
@wayfarerlabs
We're hiring. https://tally.so/r/mJxjYX
Turbocharge speed & quality gains in Diffusion World Models! 🚨 ︀︀ ︀︀- 8x8 AE w/ depth latents → 4x fewer tokens, 4x FPS boost ︀︀- 4x4 flow+depth AE in progress → next-level consistency ︀︀- DMD distillation: 16→2 steps = 8x faster sampling ︀︀- Custom RoPE fix → 20x…
We're skipping the blog post this week since all of our team is at ICML. We're excited to share what we've been working on next week :)
Anyone at ICML wanna see the future of world models? We're walking around with a laptop running our world model at 500 FPS+ (fully local). Would love to demo/chat with anyone interested
Gonna be at ICML Monday lu.ma/hmnk49rg if anyone is in town lmk
What!? A Tuesday release from Wayfarer Labs?! Heresy! Good thing the post is so good :)
Turbocharge speed & quality gains in Diffusion World Models! 🚨 ︀︀ ︀︀- 8x8 AE w/ depth latents → 4x fewer tokens, 4x FPS boost ︀︀- 4x4 flow+depth AE in progress → next-level consistency ︀︀- DMD distillation: 16→2 steps = 8x faster sampling ︀︀- Custom RoPE fix → 20x…
We're delaying our release this week a few days because most of our team is busy at an offsite today :) excited to share what we're working on.
Every single week we're posting an update. Fridays. Don't blink. 👁️
How far can we push visual fidelity with 4x4 latents? ✨ GANs help, just not how you'd expect. 🔹 Decoupled reconstruction + generation 🔹 ControlNet photorealistic supervision 🔹 100+ fps at 360p 🔹 Autoencoder + GAN → unlocked diffusion Blog in comments 👀 1/3
FYI, we're rebranding this week. Stay tuned to see our new coat of paint :)